Flyte enables production-grade orchestration for machine learning workflows and data processing created to accelerate local workflows to production.

Flyte

Hi all,
2 questions:
1. When running a workflow locally, is there a way to run tasks in parallel (for example in a map_task) or all tasks essentially have to be ran in sequence? It would be very valuable to use all the available cores.
2. When using a dataclass as a task input, is it possible to type hint a generic type e.g BaseDataclass, and then pass a class that inherits from it e.g AdvancedDataclass? I'm encountering some issues with that

1. Not today, it can be done. We have a lot of work happening in async support which will make this possible - but not planned right now. You can contribute
2. what issues? This should work

1. I see. I will consider that.
2. yes it works

thanks ketan

<@UNZB4NW3S>
what I meant in 2 is that if I have a workflow that takes a base_class type as input (the actual passed type is advanced_class), then passes it to a task, the task will receive a base_class object instead of advanced_class, with some information missing.
isnt it the case?

image.png

error when workflow receives baseclass and task receives advanced class